How well do CBT-I trials report adverse events compared with medication trials?

CBT-I trials tend to report safety casually, not systematically

In a randomized trial of CBT-I for cancer survivors, researchers specifically examined how adverse events were recorded and attributed during treatment [2]. The fact that this required a dedicated analysis suggests that adverse-event recording in CBT-I trials is not routine or standardized. The abstract does not report a comparison with medication trials, so it cannot tell us whether CBT-I trials are better or worse than medication trials at capturing harms—only that the process needed scrutiny.

Smaller CBT-I studies often report safety in a single sentence. A pilot study of brief CBT-I delivered by a nurse in Japan stated that there were no dropouts and no adverse events among nine patients [4]. That is a reassuring signal, but with nine participants it is a weak basis for judging safety, and the study did not compare its safety reporting with any medication trial.

Why a fair comparison with medication trials is hard to make

Medication trials typically have structured safety protocols: predefined adverse events, regular monitoring, and systematic attribution of symptoms to the drug. CBT-I trials, by contrast, often treat adverse events as a secondary or incidental outcome. The COZI study—a multi-center randomized trial comparing digital CBT-I, zolpidem or trazodone, and their combination in rural adults—lists adverse events as a secondary outcome alongside insomnia severity [3]. That design does allow a within-trial comparison of safety between CBT-I and medication, but the paper describes the study design and recruitment rather than reporting the adverse-event results, so we do not yet know what that comparison showed.

A non-inferiority trial comparing Tai Chi with CBT-I in older adults also lists adverse events as a secondary outcome [5]. This trial compares two non-drug treatments rather than CBT-I against medication, so it can tell us about CBT-I's safety profile relative to another behavioral approach but not relative to drugs.

The safety signals that do exist point toward few harms, with caveats

Where CBT-I trials do report adverse events, the picture is mostly benign. The Japanese pilot found no adverse events in nine patients [4], and the cancer-survivor trial focused on recording and attributing adverse events during CBT-I [2]. These are consistent with CBT-I being a low-risk behavioral treatment, but the samples are small and the reporting methods vary, so the absence of reported harms should not be read as proof that no harms occur.

The SEDATIVE pilot trial adds a different kind of safety information: it tracked side effects related to reducing sleep medications in Veterans receiving digital CBT-I alongside pharmacist-led deprescribing [1]. That is not a comparison with medication trials, but it shows that when CBT-I is paired with medication reduction, adverse events from the medication taper are explicitly monitored—a level of safety attention that CBT-I-only trials often lack.

What this means if you are weighing CBT-I against sleeping pills

The honest answer is that CBT-I trials have not yet produced the kind of systematic safety data that medication trials generate. The COZI trial was designed to compare adverse events between CBT-I and medication directly [3], but its results are not reported in the available summary. Until such comparative safety data are published, the strongest statement the evidence supports is that CBT-I trials seldom report serious adverse events, while medication trials document a known side-effect profile—not that CBT-I has been proven safer in a head-to-head safety comparison.

Sources used in this answer

Related research questions